Psychological, Physiological and Psycho-Acoustical Correlates of Strong Emotion in Music
Dr. rer. nat. in the field of neuroscience
Institute for Music Physiology and Musicians’ Medicine Hanover University of Music and Drama
Center for Systems Neuroscience, Hanover (ZSN)
University of Veterinary Medicine Hanover
by Oliver Grewe
from Haltern
Hannover 2007
Supervisor: Prof. Dr. Eckart Altenmüller
1
stEvaluation: Prof. Dr. Eckart Altenmüller (University for music and drama, Hanover) 2
ndEvaluation: Prof. Dr. Elke Zimmermann
(University of Veterinary Medicine Hanover) 3
rdEvaluation: PD Dr. Karl-Heinz Esser
(University of Veterinary Medicine Hanover) 4
thEvaluation: Prof. Dr. Andreas Lehmann
(School of music, Würzburg)
Date of final exam: 26.04.07
Published contents:
Grewe, O., Nagel, F., Kopiez, R., & Altenmüller, E. (2007). Listening to music as a re-creative process - Physiological, psychological and
psychoacoustical correlates of chills and strong emotions. Music Perception, 24(3).
Grewe, O., Nagel, F., Kopiez, R., & Altenmüller, E. (in press).
Emotions over time. Synchronicity and development of subjective, physiological and mimic affective reactions to music. Emotion.
This work was supported by the DFG (Al 269-6) and the
Center for Systems Neurosciences Hanover
Summary
Psychological, Physiological and Psycho-Acoustical Correlates of Strong Emotion in Music
Oliver Grewe
Emotional reactions show dynamic relations between subjective feelings and physiological arousal. Music is an ideal stimulus to study the dynamics of emotion components because, like emotions it develops over time and can even elicit ecstatic affective “chill” reactions (goose bumps, shivers). Most people are able to identify basic emotions expressed in music and experience affective reactions to music. But does music generally induce emotion? Does it elicit subjective feelings, physiological arousal and motor reactions reliably in different individuals? It is still under debate whether these reactions are based on distinct evolutionary, and universal reaction patterns or are acquired during a process of individual acculturation. In this interdisciplinary study, measurement of physiological arousal (skin conductance, heart rate, breathing rate), motor responses (facial muscle activity), and self-monitoring were synchronized with musical stimuli. Two experiments were performed in order to investigate the emotional procession of aesthetic stimuli.
In the first experiment, a heterogeneous group of 38 participants listened to classical, rock and pop music and reported their feelings in a two-dimensional emotion space. The first entrance of a solo voice or choir and the beginning of new sections were found to elicit inter-individual changes in subjective feelings and physiological arousal. Quincy Jones’ “Bossa Nova” motivated movement and laughing in more than half of the participants. Bodily reactions such as “goose bumps” and “shivers” (Chills) could be stimulated by the “Tuba Mirum” from Mozart’s Requiem in 7 out of 38 participants. Chills showed a relation to events in music. These were the entrance of a voice, a distinct motive or theme, changes in loudness and the contrast of two voices.
Chill responders could be distinguished from non chill responders by two character items. Chill responders showed higher values on the reward dependence scale (TCI) and lower values on the thrill- and adventure seeking scale (SSS-V). Furthermore, chill responders reported to be more familiar with classical music, to identify more with the musical style they prefer, and to listen music in a context similar to the experiment.
Additionally, we repeated the experiment seven times with one participant in order to examine the intra-individual stability of the observed effects. We found that chills were stable in response to events like entry of a voice or a distinct motive for up to six days.
In the second experiment, we used the results from the first experiment to modify the selection of participants, stimuli, the self report and physiological measurement. The experimental setting could be designed according to distinct hypothesises in order to get more homogeneous data with the results of the first exploratory study. Participants belonged to two groups, one of which was highly familiar with the stimulus (five movements from the Requiem KV 626 of Mozart), the other of which was unfamiliar with the stimulus. Familiarity raised the frequency of chills and the reported intensity. The findings of the second experiment confirmed the results of the first study. Changes in felt intensity, chill frequency, and physiological arousal were elicited by, a raised physical intensity of the stimulus (loudness), the beginning of a melody or motive and the entry of a lead voice or choir. Components of emotions were found to be independent factors which were related on various levels.
Based on the findings of both experiments a model of the relations between music perception, emotional processing and affective reactions is suggested. In this model we discuss the impact of physical intensity of the stimulus, orienting responses, and raised attention while listening on affective reactions. The emotion components (subjective feeling, physiological arousal) and chills were observed over time. Thus, the dynamic aspects of emotion were discussed. This exploratory combination of approaches throws a new light on the astonishing complexity of affective music listening.
Zusammenfassung
Psychologische, physiologische und psycho-akustische Korrelate starker Emotionen in Musik
Oliver Grewe
Emotionale Reaktionen zeigen dynamische Beziehungen zwischen subjektiven Gefühlen und physiologischer Erregung. Musik stellt einen idealen Stimulus bei der Untersuchung dynamischer emotionaler Komponenten dar; sie entwickelt sich über die Zeit und kann exstatische affektive Reaktionen bis hin zu „Chills“ (Gänsehaut, Schauer über den Rücken) auslösen. Die meisten Menschen sind in der Lage durch Musik ausgedrückte Basisemotionen zu identifizieren und erleben affektive Reaktionen beim Musikhören. Doch löst Musik regelhaft Emotionen aus? Kann sie in verschiedenen Individuen zuverlässig subjektive Gefühle, physiologische Erregung und motorische Antworten induzieren? Es besteht eine aktuelle Diskussion ob diese Reaktionen auf bestimmten, evolutionären, universalen Reaktionsmuster beruhen, oder während des Prozesses individueller Entwicklung erworben werden. In dieser interdisziplinären Studie wurden physiologische Messungen (Hautwiderstand, Herzrate, Atemrate), motorische Reaktionen (Mimik) und von den Probanden angegebene Gefühle mit musikalischen Stimuli synchronisiert.
Im ersten Experiment bildeten 38 Hörer ihre Emotionen in einem zweidimensionalen Emotionenraum ab, während sie Musik hörten. Der erste Einsatz einer Solostimme oder eines Chores, sowie der Beginn eines neuen Abschnitts lösten inter-individuelle Änderungen sowohl in den subjektiven Gefühlen als auch in der physiologischen Erregung aus. Quincy Jones’ „Bossa Nova“ löste das Bedürfnis nach Bewegung und Lachen in mehr als der Hälfte der Teilnehmer aus. Körperreaktionen wie „Gänsehaut“ oder „Schauer über den Rücken“ konnten in 7 von 38 Teilnehmern durch das „Tuba Mirum“ aus Mozarts Requiem KV626 stimuliert werden. Chills zeigten über alle Stücke einen Zusammenhang mit musikalischen Ereignissen: dem Einsatz einer Stimme, eines bestimmten Motivs oder Themas, Veränderungen in der Lautheit und dem Kontrast zweier Stimmen. Probanden mit vielen Chills (im weiteren
„Chillresponder“) ließen sich anhand von zwei Charaktereigenschaften von Probanden, die keine Chills berichteten, unterscheiden. Die Chillresponder zeigten höhere Werte
auf der „Belohnungsabhängigkeits“ Skala (TCI) und geringere Werte auf der „Thrill- und Abenteuer-Suche“ Skala (SSS-V). Darüber hinaus gaben Chillresponder an, mit klassischer Musik besser vertraut zu sein, sich mehr mit der von ihnen bevorzugten Musik zu identifizieren und normalerweise in einem Kontext Musik zu hören, welcher der Situation im Experiment ähnelt. Zusätzlich wurde das Experiment an sieben aufeinander folgenden Tagen mit einer Teilnehmerin wiederholt. Dies diente der Kontrolle der individuellen Stabilität der Effekte. Chills traten dabei in Reaktion auf Ereignisse wie den Einsatz einer Stimme oder ein bestimmtes Motiv an bis zu sechs Tagen wiederholt auf.
Im zweiten Experiment wurde, basierend auf den vorigen Ergebnissen, die Auswahl der Probanden, der Stimuli, der Art der Selbstauskunft, und zum Teil die physiologischen Messungen modifiziert. Anhand der Ergebnisse der ersten Studie wurde also der experimentelle Aufbau anhand genauer Hypothesen entworfen, um Datenrauschen zu minimieren. Die Teilnehmer dieses Experiments stammten aus zwei Gruppen, von denen eine mit dem Stimulus (fünf Sätze aus dem Requiem KV626 von W.A. Mozart) hochgradig vertraut waren, während die andere Gruppe das Stück nicht kannte. Mit der Vertrautheit stiegt die Häufigkeit von Chills und die berichtete Intensität der Gefühle. Die Ergebnisse der zweiten Studie bestätigten die des ersten Experiments. Veränderungen in der gefühlten Intensität, Chillhäufigkeit und physiologische Erregung wurden durch erhöhte physikalische Intensität (Lautheit) des Stimulus, Beginn einer Melodie oder eines Motivs, bzw. Einsatz einer Solostimme oder eines Chores ausgelöst. Emotionskomponenten stellten sich als unabhängige Faktoren heraus, die auf verschiedenen Ebenen miteinander in Beziehung stehen.
Basierend auf den Ergebnissen beider Experimente wird ein Modell der Beziehungen zwischen Musikwahrnehmung, emotionaler Verarbeitung und affektiven Reaktionen vorgeschlagen. In diesem Modell wird der Einfluss von physikalischer Intensität, Orientierungsreaktionen und erhöhter Aufmerksamkeit beim Hören auf affektive Reaktionen diskutiert. Komponenten emotionaler Reaktionen (subjektives Gefühl, physiologische Erregung) und Chills wurden in ihrem zeitlichen Verlauf beobachtet, was die Betrachtung dynamischer Aspekte von Emotionen erlaubt. Diese explorative Kombination verschiedener Ansätze wirft ein neues Licht auf die erstaunliche Komplexität affektiven Musikhörens.
Contents
1. Introduction 4
2. Background 5
2.1. General introduction to the field of emotion 5
2.1.1. How can “emotion” be defined? 5
2.1.1.1. Physiological arousal 2.1.1.2. Feeling
2.1.1.3. Motor response 2.1.1.4. Behavioural component 2.1.1.5. Cognitive component
2.1.2. Utilitarian and aesthetic emotions 6 2.1.3. Music is an ideal stimulus to study emotional reactions 6 2.1.4. Universal affective reactions in response to music? 7 2.1.5. Distinguishing emotion expressed in music from 8
subjectively perceived feelings
2.1.6. Dynamic and static aspects of the description of emotion 9 2.1.6.1. Emotions can be described as states
2.1.6.2. Emotions can be described as processes
2.1.7. Studying emotion 11
2.2. Subjective self-report of feelings 11
2.2.1.1. Emotion and reason 2.2.1.2. Emotion and music 2.2.1.3. Music and language
2.2.2. Models for the rating of feelings 14 2.2.2.1. Universal fundamental emotions
2.2.2.2. Eclectic adjective lists 2.2.2.3. Dimensional models
2.3. Physiological component 16
2.4. Motor component 17
2.5. Chills as an parameter for emotional peak experiences 17 2.6. What advantages result from a multi-methodological approach 18
based on second-per-second measurement?
2.6.1. The EMuJoy software 19
2.6.2. Psychoacoustical analysis 19
2.6.3. Comparing individual and inter-individual emotional 20 reactions
3. Hypothesis and Aims of the First Experiment 21 3.1. Dynamics and synchronicity of emotion 21
3.1.1.1. Physiological arousal component 3.1.1.2. Subjective feeling component 3.1.1.3. Motor expression component
3.1.2. Hypothesis and aims regarding dynamic aspects of 23 emotion.
3.2. Chills as an indicator of strong emotion 23 3.2.1. Hypothesis and aims regarding Chills as an indicator 25
of strong emotions
4. Materials and Task of the First Experiment 27
4.1.1. Participants 27
4.1.2. Materials 27
4.1.2.1. Auditory Stimuli
4.1.2.2. International Affective Picture System
4.1.2.3. Questionnaires 4.1.2.4. Self-monitoring
4.1.3. Apparatus and Procedure 30
4.1.4. Further Data processing regarding dynamic aspects 32 of emotion
4.1.4.1. Medians of EMuJoy data
4.1.4.2. Differences in EMuJoy data (DEMJ) 4.1.4.3. Skin conductance data
4.1.4.4. Facial muscle EMG 4.1.4.5. Criterion for affective events 4.1.4.6. Rating of musical structures
4.1.5. Further data processing regarding chills as an 33 indicator of strong emotions
4.1.5.1. Chill criteria
4.1.5.2. Character inventories and music questionnaires 4.1.5.3. Participants’ statements regarding the chill pieces 4.1.5.4. Means of psychoacoustical time-series
4.1.5.5. Frequency of peaks in psychoacoustical parameters related to chills
5. Results of the First Experiment 38
5.1.Dynamics and synchronicity of emotion 38
5.1.1. Overall reactions 38
5.1.1.1. Physiological and motor component 5.1.1.2. Subjective feeling component
5.1.2. Second-per-second responses 40
5.1.2.1. Feeling component - self-monitoring in the 2-DES
5.1.3. Time series data 41
5.1.3.1. Median curves of self monitoring data
5.1.3.2. Inter-individual commonality in self-report and SCR 5.1.3.3. Musical features, self-monitoring and physiological reaction
5.1.4. Retest in one individual 46
5.2.Discussion of Dynamics and synchronicity in emotion 49
5.2.1. Overall ratings 49
5.2.1.1. Physiological arousal component 5.2.1.2. Motor component
5.2.1.3. Subjective feeling component
5.2.2. Second-per-second measurements 50
5.2.3. Does music generally induce emotions? 50 5.2.4. How can affective reactions in the single components 51
be explained then?
5.2.4.1. Physiological arousal component 5.2.4.2. Motor reaction component 5.2.4.3. Subjective feeling component 5.2.4.4. Retest experiment in one individual
5.2.5. Conclusion 53
5.3.Chills as an indicator of strong emotions 55 5.3.1. Case study: Bach “Toccata BWV 540” 55
5.3.2. Retest Experiment 58
5.3.3. Frequency of Chills 59
5.3.4. Questionnaires 60
5.3.5. Psychoacoustic Analysis 61
5.3.6. Chills and Pleasantness 63
5.4.Discussion of chills as an indicator of strong emotions 66
6. Hypothesis and Aims of the Second Experiment 72 7. Materials and Task of the Second Experiment 74
7.1.1. Participants 74
7.1.2. Materials and task 74
7.1.2.1. Self-monitoring 7.1.2.2. Self-report of chills
7.1.3. Stimuli 75
7.1.3.1. Acoustical stimuli 7.1.3.2. Visual stimuli 7.1.3.3. Apparatus
7.1.4. Procedure 78
7.1.5. Further Data processing 78
7.1.5.1. Comparison of intensity levels between the test and control groups
7.1.5.2. Changes in intensity ratings over time 7.1.5.3. Psychoacoustical loudness analysis 7.1.5.4. Skin conductance data
7.1.5.5. Normalization of heart rate 7.1.5.6. Normalization of breathing rate
8. Results of the Second Experiment 80
8.1.1. International Affective Picture System
8.1.2. Physiological reactions in response to the beginning 80 of a stimulus
8.1.3. The test group reported higher intensity and more chills 82 than the control group
8.1.3.1. Relation of loudness with intensity ratings, chills, and physiological data in response to the Lacrimosa
8.1.3.2. Comparison between the test group and the control group 8.1.3.3. Relation of loudness with intensity ratings, chills, and
physiological data in response to the Confutatis
8.1.3.4. Comparison between the test group and the control group 8.1.3.5. Relation of loudness with intensity ratings, chills, and
physiological data in response to the Tuba Mirum 8.1.3.6. Comparison between the test group and the control group
9. Discussion 91
10.References 99
11.Acknowledgements 106
Introduction
Introduction
But when the ship was within the sound of a man's shout from the land, we fleeing swiftly on our way, the Sirens espied the swift ship speeding toward them, and they raised their clear-toned song: 'hither, come hither, renowned Odysseus, great glory of the Achaeans, here stay thy barque, that thou mayest listen to the voice of us twain. For none hath ever driven by this way in his black ship, till he hath heard from our lips the voice sweet as the honeycomb, and hath had joy thereof and gone on his way the wiser.
(Homer's Odyssey, Book XII, Butcher-Lang translation)
The hero Odysseus had to block the ears of his companions with wax and make them tie him to the mast of his ship to escape the fatal song of the sirens. Is this powerful, “magical” force of music a mere myth, or can it in some way manipulate our emotions in real life? Are there components in music that provoke similar affective reactions in all listeners, independent of acculturation, context, and musical preferences? Can our moods be manipulated by the “right” music or do we actively construct our feelings using music as a kind of tool? Many studies concerned with emotions use music as an ideal stimulus to elicit affective responses. Emotions show dynamics of psychological and physiological components; conscious feelings and physiological arousal, peaking in ecstatic “chill” experiences, seem to be related in a complex system of reciprocal influence. Music develops over time and is thus an appropriate stimulus to study the dynamics of emotion over time.
Listening to music appears to be a passive pastime, but is it really that passive?
Basically there are two schools of thought about this topic: emotivists believe that music can directly elicit emotions in listeners, whereas cognitivists describe music as a stimulus that can express emotions that are only interpreted by the listener (Kivy, 1990; Krumhansl, 1997). Up to now, many studies have concentrated on the ability of music to express emotions (Balkwill & Thompson, 1999; Brown, 1981; Cunningham
& Sterling, 1988; Hevner, 1936; Terwogt & van Grinswen, 1991). Perceived emotions and felt emotions differ (Gabrielsson, 2002), and one’s own emotional experience seems to be the most important reason for listening to music (Panksepp, 1995). This study is an interdisciplinary project that attempts to approach the question: can the
Background
General introduction to the field of emotion
Several aspects that were of influence for the design of this study are discussed in this paragraph. Most methods used to study emotion, as well as the definition of emotions, are still under discussion and need further development. Thus, a thorough discussion of the definitions, background and assumptions underlying this study may avoid misinterpretations and misunderstandings regarding the results. Recently, research in the field of music and emotion has been criticised for conceptual and definitional shortcomings (Konecni, 2003; Scherer, 2004). Different disciplines do not seem to have “complementary views of the same phenomenon, but divergent views on substantively different phenomena” (Konecni, 2003, p.334). Thus, we tried to conscientiously summarize the terminological and conceptual foundation of this study.
How can “emotion” be defined?
To define of the terms “emotion” and “feeling” we used the “component pro- cess model” as presented by Scherer (2004). These definitions are based on a widely accepted model of emotions. According to the component process model the terms emotion and feeling are not used synonymously. An emotion consists of three
“classical” components:
Physiological arousal. The physiological component of an emotion represents the measurable signs of bodily stimulation or arousal. It can be caused by the disturbance of the homeostatic equilibrium by an emotional event. As a component of an emotion the physiological arousal is thought to prepare the body to some kind of fast adaptive reaction to the emotional event (fight or flight).
Feeling. The feeling component of an emotion represents the consciously perceived, felt affective event. A feeling must not necessarily correspond to an unam- biguous verbal label like “happy” or “sad”. It reflects internal states and means a moti- vational change. There seems to be aproprioceptive feedback between feelings and the physiological and motor component.
Motor response. A spontaneous motor response, such as facial movements or vocal expression is another component of an emotion. Motor responses can be a spontaneous way of communicating ones’ emotional state or reacting to an emotional stimulus. Additionally to the “emotion triad” two further components have recently been added to this definition of emotion:
Background
Behavioural component. The behavioural component of emotions consists of an interruption of ongoing goal directed behaviour, a re-evaluation of the situation and the preparation of adapted behaviour. Thus, emotions can be hypothesized to decouple stimulus and response. They lead to constant adaptation to a situation, based on feedback loops integrating the new experiences (Scherer, 1994a).
Cognitive component. The cognitive evaluation process as a component of an emotion leads to goal attainment of an individual. Deciding the impact of the emotional event on the well-being of the individual can have a strong influence on attention, problem solving, judgement and decision making. The cognitive appraisal of an emotional situation also has a recursive influence on feelings when, for example, a situation is evaluated as dangerous, it will most probably generate a feeling of fear.
Utilitarian and aesthetic emotions
The concept of emotion is based on the idea of an intuitive reaction pattern, which is needed to initiate an immediate adaptive behaviour in favour of the individual needs and goals of an individual (Darwin, 1965). This is a very plausible concept for everyday life situation. In case of aesthetic stimuli like music, pictures or films, the events and situations are mere fiction and there is no obvious need to adapt one’s behaviour. However, we seem to react in an emotional way to aesthetic stimuli.
Thus, a distinction between utilitarian and aesthetic emotions (Scherer, 2004) is useful when studying the affective effects of music. In contrast to utilitarian emotions, aesthetic emotions are not based on goal relevance and coping potential. Rather, aesthetic emotions can be influenced by the abstract appreciation of intrinsic quality values of a stimulus. The quality of an artistic work may arouse feelings of awe in us (Konečni, 2005), even if the topic of this specific piece of art does not particularly move us.
Music is an ideal stimulus to study emotional reactions
Music is regularly used in everyday life to influence one’s own emotions (Panksepp, 1995) and can elicit strong peak experiences, such as ecstatic chill reac- tions (Blood & Zatorre, 2001; Craig, 2005; Panksepp, 1998; Sloboda, 1991).
Moreover, music has further advantages that can be used as a scientific stimulus to elicit emotion. First of all, music develops over time, in contrast to “static” stimuli, such as pictures, thus making it possible to study the development of emotion in
response to a continuous stimulus. Music can offer many approaches to emotion: it can be used in individual sessions, because people frequently listen to music alone (e.g., via headphones). Music is also listened to in social situations like concerts or discotheques. Music is frequently performed alone as well as with others (e.g., in choirs). Moreover, it is often used as mere acoustic background (e.g., in shopping malls), but is also listened to attentively. Thus, music can be adapted in different situations without being perceived as unnatural or artificial. Since studying emotions is very demanding and it is absolutely necessary that participants feel comfortable and natural in the given situation, this flexibility is an immense advantage.
Universal affective reactions in response to music?
It was said before that emotions are thought to have relevance for adaptive behaviour and thus impact the evolutionary fitness of the individual. When discussing the evolutionary roots of the perception of emotions in music, the findings of inter- cultural comparisons, the reactions of children to music, and comparative studies (animal studies) become of great interest. If components of decoding emotions in music are based on our genetic inheritance and thus could be assumed to be evolutionary relevant factors, they should be found in other cultures and in early stages of human development. Abilities that are evolutionarily relevant for us may also be found in related species. The distinction between harmonic and dissonant sounds, for example, may be hypothesized to be a feature that we share with other species. The question of inter-cultural agreement of emotions coded in music was addressed by Gregory & Varney (1996). They compared the perception of Eastern (Indian) and western classical music in listeners with either an Indian or western cultural background. Judgements were given via a forced choice emotional adjective list. The comparison of the groups revealed an overall similarity in the emotion judgements; however, specific pieces showed differences. Terwogt and van Grinsven (1991) addressed the question of how the perception of basic emotions in music changes during development. They asked three age groups (5- and 10-year old children as well as adults) to link selected musical excerpts to facial expressions representing basic emotions. In this study, excerpts from exclusively western classical music were used as stimuli. Even the young children agreed considerably in their choices of expressed emotions. However, an increase of consensus in older listeners could be seen, indicating the relevance of learning processes for the decoding of
Background
emotional cues in music. Participants had more problems identifying negative emotions, such as fear and anger compared to positive emotions.
The general consensus, even in young children, regarding the emotional content of music, suggests that the perception of music may be based on some very basic principles that may even be found in the sensation of animals. McDermott and Hauser (2004) investigated whether cotton-top tamarins, often used for tests of perceptual abilities, show preferences for consonant sounds. The tamarins were placed in a v-shaped maze, which consisted of two arms with different auditory environments. Preference for one sound was measured by the time the tamarin spent in each arm of the maze. The animals differed neither in their preferences for consonant or dissonant sounds nor for screeching sounds compared to amplitude- matched white noise. Humans, in this experiment, showed clear preferences for consonant intervals over the amplitude-matched white noise. The authors conclude that the preferences that support our capacity for music could be due to music-specific adaptations.
Our study investigated individual as well as general affective reactions to music. We avoided the term “universal”, because we investigated a group of listeners highly familiar with western music. However, possible “general” reactions found in this group could be a base for inter-cultural comparisons. In particular, chills were hypothesized to be evolutionary inherited reactions to distinct acoustical pattern (Panksepp & Bernatzky, 2002).
Distinguishing emotion expressed in music from subjectively perceived feelings When we ask whether music induces emotion, it is important to distinguish between emotion expressed by the music and feelings subjectively perceived by the listener. The interpretation of the listener as to the intended expression of a piece of music may differ from the feeling the listener actually has in response to the music (Gabrielsson, 2002). When we compare physiological and psychological reactions in response to music we refer to affective responses elicited or induced in the listener.
We differentiate between “eliciting” and “inducing” because music may “induce”
emotions in a reflex like manner, independent of social and cultural background. It may also “elicit” feelings, i.e., influence the process of cognitive appraisal as one factor next to individual factors such as personal associations, familiarity with a particular musical style, mood, wakefulness, etc. One of the aims of this study was to
find a more accurate analysis of these emotionally influencing effects of music. The question whether music induces emotions is part of an intensive and ongoing debate in the literature (Juslin & Sloboda, 2001; Kivy, 1990, 2002; Konečni, 2005; Meyer, 2001).
Dynamic and static aspects of the description of emotion
Emotions can be described as states as well as processes. Naming is possible for states, since naming requires constancy (Meyer, 2001). Expressions like “sadness”
or “happiness” imply states. Some authors hypothesize that specific bodily states for every emotion exist (Ekman & Davidson, 1994; Panksepp, 1998). However, reviews of larger compilations of available studies that contrast two or more emotions (Cacioppo, Klein, Berntson, & Hatfield, 1993) found little evidence for an emotion- specific physiology. Authors like Meyer (2001) and Scherer (1993) suggest concepts of dynamic emotional processes, that occur via a synchronization of emotional subsystems, connected by feedback and forward processing (Scherer, 1993), or a combination of native and syntactic processes (Meyer, 2001).
Emotions can be described as states. Single terms for emotions refer to states and can hardly describe the dynamics involved in the processes. Terms are only an approximation and abstraction of real events (Meyer, 2001). This classification and abstraction of emotions makes terms indispensable in discussion about emotions. The expression “a state of being moved” suggested by Konečni (2005) demonstrates the
“tension” between the stationary character of terms and the dynamic character of emotions. It could be shown in several studies that the classifications of basic emotions expressed in music by different participants works reliably (Cunningham &
Sterling, 1988; Ekman & Friesen, 1971; Hevner, 1936; Terwogt & van Grinswen, 1991). However, the specificity of recognition is limited (Brown, 1981). The stimuli Brown used expressed six degrees of sadness (e.g., sadness tinged with romantic mystery). Brown could show that neither musicians nor non-musicians categorized the excerpts exactly. The qualities of music as a language for emotion will be discussed in more detail below. Here we would like to emphasize the limited accuracy of categorizing emotions expressed in music. This gives rise to the question: are only the recognition and naming of emotions limited in specificity, or is this also the case for experiencing emotions as a reaction to musical stimuli? “Given the lack of real-life observational data, we do not know how invariant or variable the appraisal, expression
Background
and labelling patterns really are. I bet that they are quite variable” (Scherer, 1994b, pp.26-27).
”Flexibility” in response to stimuli may be an important evolutionary advantage for human beings. Scherer hypothesizes that it is the flexibility in possible reactions to a stimulus that makes human the most emotional animals. “What most significantly shaped human behaviour and gave rise to human cultures was not the presence, but the absence of adequate innate constraints” (Meyer, 2001, pp. 348-349).
This study investigates how consistently emotions can be “induced” in listeners via music. That means that we did not investigate the accuracy of the ability of the listeners to label musical expressions with emotional terms, but we were looking for similarities between listeners’ emotional reactions in all components (subjective feeling, physiological arousal, motor response). When we are interested in the processing of a stimulus, the dynamics of reaction patterns becomes interesting, i.e., the process of an emotional episode. The question of the dynamics of emotions has several methodological and terminological implications that will be addressed in the following.
Emotions can be described as processes. Processes are caused by events, and thus the first question would be, what characterizes an event that evokes an emotional process? The second question would be: what could be the meaning of these processes and why do they occur?
Emotions can be thought of as evaluation processes. There seem to be two general preconditions for an event that causes an evaluation process: novelty and significance. An event that is known to us does not require evaluation, neither does an event that is irrelevant. However, significance seems to be an evaluation process complementary to the novelty evaluation. Novelty can be described as negatively related to the degree of match between the input and previous events. Significance, in contrast, seems to be positively related to the match between the input and the neuronal representations of previous relevant results (Ben-Shakhar, 1994). On the level of music perception, this seeming contradiction could be resolved as follows:
uncertainty (novelty) gives rise to emotional processes (Meyer, 1956). However, novelty in music implicates also certain knowledge about the “usual” patterns. When no underlying pattern can be found or at least presumed, uncertainty cannot be contrasted with certainty. This explains why the raise of attention requires both significance and novelty. The number of possible following musical events must be
limited because an evaluation process is based on implications and has the goal to predict future events. Without any underlying pattern, implications become impossible and an evaluation process frustrating (Meyer, 2001).
This leads to some further implications. It could be concluded that the starting point of theses processes contains an individual component, because the presented concept of significance is based on the knowledge of pattern. Pattern could be innate or learned. To give an example: a loud scream could be of significance to us either because the relevance of this acoustical structure is an innate, universal signal for danger, or because we learned early on that screams are a signal (symbol) for danger.
Meyer distinguishes between native and syntactic processes. Native processes are more general, syntactic processes rely on individual experience with musical syntax and are related to cultural and historical contexts (Meyer, 2001). The question that arises for this study is therefore: can affective reactions to music better be described as native (innate) or syntactic (learned) processes?
Studying emotion
The concept of emotion presented here leads to several points that a study about this topic has to take into consideration. First, an emotion consists of different components, each of which should be methodologically addressed. Furthermore, the static as well as dynamic aspects of emotions should be taken into consideration. In the following, we will discuss possible measurements for each component and aspects that may influence the conclusions that can be drawn from the results. It should be mentioned that our definition of emotions does not pretend to be the only one. Up to now, there is no general consensus about one definition of emotion. However, we tried to give our empirical experiments a valid and widely acknowledged theoretical background.
Subjective self-report of feelings
When psychologists want to make valid statements about subjective affective states they always rely on the self-monitoring of the participants (Russell, 1997). To date, there is no reliable objective way of measuring emotion or feelings independent of the participants’ descriptions of what they perceive. It could be argued that emotion does not require scientific research, since common sense tells us enough about our emotions. Why do we need objective methods to verify emotions? Why do we not
Background
limit our research to simply asking people what they feel? We should be aware that feelings participants report can be strongly influenced by their opinions about emotions. There are many old concepts in popular opinions regarding the relation of emotions and cognition, of music and emotion as well as the relation between music and language. These concepts are in part based on the most ancient mythological views. This study could only touch upon some aspects of psychological aesthetics. For a more detailed introduction to this field, see Allesch (1987; 2006).
Emotion and reason. The Cartesian idea that body and mind are completely separated has been questioned in the past decades (Damasio, 1994). However, “for many, Descartes’ views are regarded as self-evident and in no need of re- examination” (Damasio, 1994, p.250). The idea of emotion as body-bound phenomena separated from a reasonable mind may still influence the self-report of feelings. Thus, it is a relevant question, what aspects of an emotional episode can actually be measured as a body function. What is the impact of physiological arousal for our subjective feelings? If such stable relations can be found, do they represent individual, general or even universal principles?
Emotion and music. The idea of a division between light and clear intellectual forces (reason), and dark, uncontrolled forces of sensitivity and emotions (body), later postulated by Descartes (1998 [originally published in 1637]), can also be found on the earliest stage of western musical culture. The greek music-theoretical concepts and systems were mediated by Boetius (~480-524) in his writing De institutione musica to the Medieval period and became the base for early western music culture. The ancient theories about music go far beyond the mere acoustical phenomenon. For example, Boethius divided music into the musica instrumentalis, musica humana and musica mundana. Music as an acoustical phenomenon (musica instrumentalis) was understood as a representation of the harmony of the human body and soul (musica humana), as well as of the harmony of the [geocentric] universe (musica mundana).
Later in the Medieval period this mystic-philosophical system was interpreted in terms of Christian religion (Wagner, 1952).
In the ancient greek theory, music was strongly connected to lyrics, dance, education, sport, medicine, philosophy and mythology. The theories of Pythagoras still represent the basis of western music. His postulation of the physical principles of the intervals were also a form of a number mysticism (Allesch, 2006). Another mythical-psychological concept of strong historical influence was the division in two
principles, represented by the sons of Zeus, Apollo and Dionysus. Apollo was the god of light, clear order, and the leader of choirs of the muses (Apollo Musagetes). His antithesis, Dionysus, was the god of the intoxicating elemental forces of nature, the god of wine, dance and theatre. The knowledge, and even more beliefs about ancient Greek culture has influenced the composition, understanding and interpretation of music ever since. Whereas the Greek theories were integrated in religious ideas during the Medieval time, the Renaissance period tried to bring these ideas back to real musical life. They were always present in music culture (Wegner, 1965). The question remains whether these ideas are a truth so basic that they were found even in the earliest periods of culture, or on the contrary, the early mythological stories have influenced (and maybe misled) our interpretation of emotions and music ever since.
Music and language. In its early stages, music was always connected to lyrics, in the greek cultures as well as in early christianitiy (gregorianic chants). Polyphony was developed in church music, singing chorals, psalms and motettes, popular music was spread by story telling troubadours and instruments were mostly used for accompaniment. Instrumental music did not really start before the 15th and 16th centuries. Rhythmical patterns of early music are based on metres like dactyls and jambs (Möller & Stephan, 1991). This demonstrates how close music and language were for a long time. In our days music has often been called the language of emotion (Gabrielsson & Lindström, 1993). The ideas about music and language may also be influenced by several myths. For example, the legend of Orpheus (Ovid &
Breitenbach [Translator], 1971), who descended to the underworld to bring his dead wife Eurydice back to life. He entranced Persephone and Hades by his songs, persuading them to help him attain his goal. Music is often expected to work as such a universal language of feelings, which can induce compassion or other distinct feelings. Several previously cited studies brought evidence for music as a language of feelings; identification of basic emotions (e.g., happiness, sadness, anger) in music was found to be reliable in experts as well as in non-experts (Cunningham & Sterling, 1988; Hevner, 1936; Terwogt & van Grinswen, 1991). However, the distinction of emotions expressed by music seems to be limited to basic emotions. Brown (1981) showed that musical experts as well as non-experts fail to categorize emotions with more subtle shades of expression. Familiarity with the musical genre of the stimulus raised the inter-individual consistency of the ratings. Since the consistency of the interpretation of emotions expressed in music seems limited, the question remains:
Background
how consistent are emotional feeling reactions to music? Can distinct musical patterns induce general (or even universal) affective reaction patterns?
Models for the rating of feelings
There are three major models for the assessment and rating of emotion and feelings. Here we would like to give a brief introduction to discuss advantages as well as shortcomings of all three approaches.
Universal fundamental emotions. The model of fundamental emotions is mainly based on the assumption that individual emotions are a blend of a limited number of evolutionary, continuous, and fundamental emotions (Ekman & Davidson, 1994; Plutchik, 1980). The model of fundamental emotions goes back to the theories of Charles Darwin (1965). A study of Ekman and Friesen (1971) brought evidence that facial expression of fundamental emotion like “anger” or “happiness” can be decoded universally and is independent of culture. For the rating of aesthetic feelings the model of fundamental emotions is not appropriate, since it is unlikely that music can elicit strong and unambiguous responses, such as “fear” or “disgust”. The model of fundamental emotions has been developed and is used mostly in the field of utilitarian emotion (Scherer, 2004).
Eclectic adjective lists. An approach which is commonly used in research concerned with feelings and music are eclectic adjective lists or other semantic labels (Hevner, 1936; Krumhansl, 1997). Lists of verbal expressions of emotion can be chosen according to the aim of the study and are thereby more adapted to express the extreme richness of possible affective reactions to music. Verbal labels more likely reflect the “qualia” of feelings and should be easily understood by participants, since it is an “every day task” to express feelings with the help of words. On the other hand, an individual list of labels makes the comparability of results difficult, or even impossible, and the reliability of every new list can be doubtful. Do the chosen labels really cover all the possibilities? Moreover, a continuous rating during the listening of a musical piece is impossible using forced choice adjective lists.
Dimensional models. The possibility of continuous self-report is one of the major advantages of the third model for rating feelings, the dimensional model. The tradition of searching for a limited number of dimensions underlying all possible affective responses is founded on the work of Wilhelm Wundt (1905). Wundt suggested distinguishing between three dimensions of feelings:
pleasantness/unpleasantness, rest/activation, tension/relaxation. More recent approaches have limited the model to the use of the two dimensions valence and arousal, which cover most common feelings (Russell, 1978, 1979, 1980). Valence ranges from positive to negative valence and reflects the qualitative appraisal of the stimulus. Negative valence was defined as a feeling that is unpleasant (“unangenehm”), that participants would like to avoid and that they would like to stop.
Positive valence was defined as a pleasant feeling (“angenehm”) that participants would seek and that they would like to continue. The arousal axis was defined as ranging from calm (“beruhigt”) to arousing (“aufgeregt”). The so called “circumplex model of affect” consists in a two dimensional emotion space (2-DES), which offers substantial opportunities for the rating of feelings (Plutchik & Conte, 1997) and was frequently used in studies concerned with the emotional expression of music (Rickard
& Ritossa, 2004; Schubert, 2001, 2004; Schubert & Dunsmuir, 1999; Witvliet &
Vrana, 1995).
The concept of this model is easily understood by participants. The critique expressed by Scherer (2004) that participants have to perform a “mental principal component analysis” may, on the contrary, be interpreted as one of the advantages of the 2-DES. Of course, people do not spontaneously express their feelings in categories of valence and arousal. However, feelings in response to music are rarely covered by one or two explicit verbal labels, such as “happy”, “satisfied” or “gloomy”. The continuous shadings between positive and negative valence and arousal may reflect the actual feelings better than words.
The most convincing advantage of the 2-DES is that it is the only model that allows a continuous self-report of virtually all possible feelings during music listening.
Ratings given in the 2-DES are comparable with other data obtained in the same way, they allow the study of the development of feelings over time, detecting emotional events and observing possible synchronicity in emotion components. This last point, the relation of feelings, motor responses and physiological reactions led to the development of the EMuJoy software (Nagel, Kopiez, Grewe, & Altenmüller, in press) that represents the methodological base of this study.
Background
Physiological component
Berlyne (1971) hypothesized in his psychobiological arousal theory a distinct relation between the arousal potential of a stimulus and aesthetic responses. In the course of investigating affective responses to music, many bodily reactions, such as shivers, tears, a lump in the throat, heart palpitation, and so forth, have been observed.
Those reactions can be reported by participants and be objectively controlled in part by peripheral physiology measurements. Physiological arousal is a component of an emotion that can be measured objectively. It reflects bodily reactions independent of the interpretation of the participants. Furthermore, the physiological reaction cannot be influenced in a distinct manner by the participants. Of course, the measurement of single effects mediated by the peripheral nervous system (PNS) gives only a limited insight in the physiological arousal of a person. However, the measurement of skin conductance response (SCR) has been especially successful in former studies, where it was used to measure physiological arousal during music listening (Khalfa, Peretz, Blondin, & Robert, 2002; Krumhansl, 1997; Rickard, 2002; Witvliet & Vrana, 1995).
The SCR is the phasic part of the skin conductance level (SCL). The SCL is mainly based on sweating reactions and changes in blood flow (Boucsein, 2001). These reactions have been widely used as indicators of psychological reactivity, especially as an indicator of arousal, orienting responses and startle responses. SCR reflects the changes in SCL over a short period and is therefore adapted to measure reactions to affective musical events. Krumhansl (1997) recorded a wide spectrum of physiological measurements of cardiac, vascular, electrodermal, and respiratory functions as well as emotion quality ratings (fear, sadness, happiness) on a second- per-second basis and correlated the averages of the measurements. She found the strongest correlations at the skin conductance level (SCL), increasing significantly during all three reported emotion qualities. Rickard (2002) found SCR to be the most effective of all tested physiological parameters (heart rate, skin temperature, EMG, level of cortisol) to measure physiological arousal. SCL and SCR are easy to obtain measurements compared to the level of cortisol. Furthermore, the SCR reflects the previously mentioned orientation response, which was hypothesized as a starting point for emotional processes. Additionally, SCR can be used as an objective indicator of chill responses. As discussed later, the activation of the sympathic nervous system that
leads to goose bumps and shivers should also lead to measurable responses in SCR and SCL (Craig, 2005).
Motor component
Gaining information about specific motor responses is not an easy task, especially when physiological arousal is being measured at the same time. The movement of the whole body does, of course, influence sweating reactions and blood circulation. Furthermore, participants cannot dance in an experimental setting, one of the most natural motor responses to music. However, measuring EMG responses of facial muscles has shown to be an interesting emotion-related measure of motor components. Facial movement reactions to emotional events are one of the most intuitive and universal responses (Ekman & Friesen, 1971; Witvliet & Vrana, 1995).
Thus, we measured the EMG of the m.zygomaticus and the m.corrugator. Activity of the zygomatic muscle indicates positive valence (smiling, laughing), whereas activity of the corrugator can be interpreted as a spontaneous reaction of negative valence (disgust, aggression, frowning). Activity of facial muscles should not influence SC- reactions measured at the hand and do not disturb music listening in any way.
Additionally, in questionnaires we asked for the wish for movement. This self- reported desire could be interpreted as an indicator of motivation for movement.
Chills as an parameter for emotional peak experiences
In 1980, Goldstein (1980) published a study based on questionnaires, which for the first time used “thrills,” or “chills” as the phenomenon was called in later publications, as a parameter for strong emotions in response to music. Except in cases where a cited author used the term “thrill,” from here on we will use the term “chills”
according to Panksepp’s discussion of the terms (Panksepp, 1995). Chills are a subtle nervous tremor caused by intense emotion. Goldstein used them as an indicator for strong emotional responses, allowing the combination of psychological self-report with distinct bodily reactions such as “goose bumps” or shivers down the spine.
Goldstein applied music as a stimulus in order to test the hypothesis that thrills (chills), as an example of emotional reactions, are mediated by endorphins. He studied the effect of the opiate antagonist naloxone on the frequency, duration, and intensity of thrills and found some evidence that thrills may be attenuated by naloxone.
Interestingly, not all participants were susceptible to thrills. Out of the three groups of participants, 10% of the music students, 20% of the medical students, and 47% of the
Background
employees of an addiction research center responded in the questionnaire that they had never experienced thrills.
The “chill” parameter represents an outstanding concept to study strong emotional reactions to music. In this study, we defined chills as “goose bumps” or
“shivers down the spine”, since these appear to be the most frequent chill reactions in response to music (Sloboda, 1991). The chill paradigm offers several advantages when one is interested in strong emotions. First, it can be easily explained to the participant. Instead of agreeing to some kind of a “strong emotion definition”, the participant can simply be asked to press a button whenever the bodily response occurs.
Second, a chill has a clear beginning and end, whereas it is extremely difficult for a participant to decide when a “normal” emotion becomes a “strong” emotion. Third, goose bumps and shivers down the spine are both mediated by the sympathic nervous system. As mentioned before, activation of the sympaticus is reflected in the Skin Conductance Level (SCL) (Craig, 2005) and Skin Conductance Response (SCR), which will be addressed in more detail in the following. Thus, recording the SCR provides a useful and objective measurement for the validation of reported chill responses.
What advantages result from a multi-methodological approach based on second-per- second measurement?
In the previous paragraphs we summarized several traditional views of emotion that might influence the subjective self-report of feelings. It appears useful to supplement subjective methods of self monitoring with more objective measurements.
Including physiological measurements synchronized to self-reported feeling allows us to look at the relation between conscious statements regarding feelings and bodily responses. Moreover, the definition of emotion we based our research on asks for measurements of physiological arousal and motor activity in response to the stimulus.
According to Scherer’s definition of an emotion, the self-report of participants provides us with information about feelings. However, feelings and emotions are not the same. Using self-reports, physiological arousal, and facial motor responses in a synchronized second-per-second paradigm allowed us to test for a possible synchronization between these factors, as hypothesized by Scherer (2004).
Additionally, we included the chill parameter as an indicator for peak experiences. We
developed an approach to measure all three emotion components on a second-per- second basis in our study.
The EMuJoy software
The EMuJoy was developed to give participants the opportunity to report their feelings continuously while listening to music. EMuJoy is based on the circumplex model of affect presented by Russell in 1980 (Russell, 1980). Participants move a cursor in a two-dimensional emotion space (2-DES) on the axes valence and arousal.
The cursor in EMuJoy can be moved with a computer mouse or joystick. A joystick offers the advantage that the cursor is centred automatically, allowing the use of EMuJoy with closed eyes. However, participants preferred the mouse. This gave a hint to the specific use of EMuJoy: participants often chose a position in the 2-DES and remained there until, related to specific events in music, they changed their position. When using the joystick as input device, remaining at a specific point in the 2-DES requires constant work against the centring springs of the joystick. When using a mouse as the input device, remaining at one point requires no force at all. This specific use of the self-report device will be discussed later on.
We explicitly asked the participants to report their own feelings in response to the music rather than the emotions they think the music is intended to express. We were interested whether music can induce emotion and wanted to compare self-report with objectively measurable physiological reactions, which were expected to be related to perceived feelings. Therefore, the data from EMuJoy was synchronized with the physiological measurements in the range of milliseconds.
Psychoacoustical analysis
Panksepp (1998) hypothesized that chills may be generally related to distinct acoustical pattern and represent an innate reaction to genetically fixed acoustical reaction patterns. In order to control for this hypothesis we analysed the acoustical structure of 190 musical excerpts that were found to elicit chills in the listeners.
Loudness, sharpness, roughness and fluctuation were analysed as an average over all excerpts, as well as in various categories based on gender, positive and negative personal memories, different levels of familiarity with the stimulus, etc. Additionally, the acoustical features of chill eliciting excerpts were compared to those of random excerpts that did not elicit chills.
Background
Comparing individual and inter-individual emotional reactions
In order to investigate the general affective reactions, a heterogeneous group of listeners was tested on musical pieces from various styles. Additionally, the reactions of one listener in repeated sessions were recorded in an exploratory approach. This experiment was performed to get information about the stability of reactions and to compare these to possible general responses. Possible influencing factors such as musical background, personal interest, general physiological reactivity and personality can be assumed as stable. Thus, these results could be valuable information complementary to the data we collected for a group of listeners.
Hypothesis and Aims of the First Experiment
Dynamics and synchronicity of emotion
The components of the preceding emotion response triad require special methodological approaches balanced and coordinated with one another. We hope to overcome some of the limitations of each single method by using a conscientious combination. For each component we chose one approach to examine the overall reaction to a whole musical piece as well as second-by-second measurements. This enabled us to examine
1. responses to a piece as a whole (these responses will be called the overall condition in the following)
2. distinct musical events causing changes in affective reactions (these responses will be called the second-per-second condition)
3. possible synchronicity in the above mentioned emotion components (second- per-second condition).
Overall as well as second-per-second conditions are structured according to the emotion response triad of physiological arousal, motor response, and subjective feeling.
Physiological arousal component. Because we were interested in the effect of distinct musical events as well as possible synchronicity in affective reactions, we decided on a second-by-second measurement of skin conductance response (SCR) as an easily obtainable and reliable indicator of physiological arousal. Even if there is no emotion specific physiology, according to the component process model a physiological reaction is a basic component of emotion. The model reflects the phenomenon of a non-specific, diffuse physiological reaction in the distinction between aesthetic and utilitarian emotions. The way the affective reaction is interpreted (i.e., which specific emotion is perceived) can be understood using the subjective feeling component.
Subjective feeling component. There are three major theoretical approaches to describe participants’ conscious experiences while listening to music: discrete emotion theory, dimension models, and eclectic approaches (Scherer, 2004). In this study we chose the two-dimension emotion space (2-DES) model developed by Russell (1980). Schubert (2004) used this model and found relations between ratings of valence and arousal and the acoustical structure of musical pieces. The affective ratings referred to the interpretation of music but not to the personal affective
Hypothesis and Aims of the First Experiment
response. Major arguments against studies that rely exclusively on dimensional models are that they ask for feelings instead of emotions; that is, they control for just one component of emotions (Scherer, 2004). To solve this problem in the present study we used the model exactly for this purpose and added physiological measurements for the remaining emotion components. Second, the possibility of continuous measurement of feelings does not mean that participants constantly use the input device to express their feelings (Konečni, 2003). A computer mouse or slider may be moved just a couple of times or never while listening to a piece of music. To our knowledge, there is still no valid theory of how often affective reactions change over time. Thus, in this article the second-per-second data is presented as a time series. Even when participants altered their positions in the 2-DES just at distinct times, this could also be used as a source of information. We thought it would be interesting to analyze when the participants altered their reported subjective feelings and whether they did so in response to distinct musical events. This revealed the dynamics between perceived affective states. Thus, we additionally calculated the differentiation of the second-by-second data to gain this information about dynamic processes within the reported self-monitoring.
Third, it has been discussed above whether people are capable of expressing their feelings properly on just two axes (valence and arousal), because they are constrained to perform a mental principle component analysis (Scherer, 2004) to
“target” the feelings they want to express. The underlying evaluative mechanisms cannot be traced using this method. This is one of the most serious problems using dimensional models which cannot be solved even today to full satisfaction.
On the other hand, dimensional models offer the representation of many different emotions: they are reliable and economical and allow for second-by-second measurement. Being aware of the preceding problems, we nevertheless found the 2- DES to be the most adaptive tool to meet our intention of identifying distinct musical events and synchronicity of affective responses. We modified the approach of Schubert (Schubert & Dunsmuir, 1999) by asking for felt, instead of perceived, emotions (Gabrielsson, 2002). Participants were explicitly asked to concentrate on their own feelings and not to rate the expression of emotions. For the overall rating, we used additional questionnaires after each piece of music.
Motor expression component. In a music-listening experiment which uses physiological measurements, checking for the motor expression component becomes a difficult task. Participants were asked to sit in an armchair so that their possibilities to move were highly constrained. To take into account the restricted mobility during the experiment, we included the “wish for movement” in the questionnaires. In the event that the music was intended to motivate motor expression, this was made known to the listener, especially when motor expression was suppressed due to the experimental setting.
Another possibility to control for spontaneous motor reactions to music is to measure facial muscle responses (Witvliet & Vrana, 1995). Witvliet and Vrana (1995) examined facial electromyographic (EMG) data collected at the zygomatic (smiling) and corrugator (frowning) muscles in relation to the dimensions valence and arousal.
Zygomatic muscle tension was significantly greater during positive than during negative valence affects and greater during high arousal than during low arousal contexts. In contrast, corrugator activity was higher in negative valence and low arousal contexts. Mimic reactions are part of a typical emotional reaction pattern and seem to be universal (Ekman & Davidson, 1994). Thus, we decided to record the electromyogram of zygomatic and corrugator muscles for the second-by-second measurements.
Hypothesis and aims regarding dynamic aspects of emotion
Based on the cited literature we drew up the hypothesis that distinct musical patterns can generally induce emotions. In previous studies, music was shown to affect three major emotion components: physiological change (Krumhansl, 1997;
Sloboda, 1991; Witvliet & Vrana, 1995), motor activation (Witvliet & Vrana, 1995), and subjective feelings (Krumhansl, 1997; Schubert, 2004). Our aim was to describe the capacity of entire musical pieces to alter the overall emotional state (feelings, bodily reactions, motor activation) and to identify distinct musical events that induce significant changes in all three emotion components.
Chills as an indicator of strong emotions
Chills can be a fascinating phenomenon to study strong emotional reactions to aesthetic stimuli since they allow for verification of subjective reports with physiological measurements. They are considered to be distinct events, and therefore,
